Modelling Sensory Attenuation as Bayesian Causal Inference across two Datasets

• DOI: 10.31234/osf.io/u5fsj
• Publication Date: 2024-03-27T11:46:47Z
• Authors: Anna-Lena Eckert; Elena Fuehrer; Christina Victori...; Benjamin Straube; Katja Fiehler; Dominik Endres
• Keywords: Causal model

Introduction. To interact with the environment, it is crucial to distinguish between sensory information that externally generated and inputs are self-generated. The consequences of one’s own movements tend induce attenuated behavioral- neural responses compared inputs. We propose a computational model attenuation (SA) based on Bayesian Causal Inference, where SA occurs when an internal cause for inferred. Methods. Experiment 1investigates during stroking movement. Tactile stimuli finger were suppressed, especially they predictable. 2 showed impaired delay detection arm movement video participants moving vs. their was moved passively. reconsider these results from perspective Inference (BCI). Using hierarchical Markov Model (HMM) variational message passing, we first qualitatively capture patterns task behavior in simulations. Next, identify participant-specific parameters both experiments using optimization. Results. A sequential BCI well equipped empirical across datasets. optimized parameters, find good agreement data predictions, capturing tactile detections 1 2. Discussion. appropriate framework humans. Computational models may help bridge gap different modalities experimental paradigms contribute towards improved description understanding deficits specific patient groups (e.g. schizophrenia).